v0.50.0
Browse filesSee https://github.com/qualcomm/ai-hub-models/releases/v0.50.0 for changelog.
- README.md +65 -65
- release_assets.json +1 -0
README.md
CHANGED
|
@@ -15,7 +15,7 @@ pipeline_tag: automatic-speech-recognition
|
|
| 15 |
We have applied w8a16 quantization to significantly enhance performance and efficiency. HuggingFace Whisper-Small ASR (Automatic Speech Recognition) model is a state-of-the-art system designed for transcribing spoken language into written text. This model is based on the transformer architecture and has been optimized for edge inference by replacing Multi-Head Attention (MHA) with Single-Head Attention (SHA) and linear layers with convolutional (conv) layers. It exhibits robust performance in realistic, noisy environments, making it highly reliable for real-world applications. Specifically, it excels in long-form transcription, capable of accurately transcribing audio clips up to 30 seconds long. Time to the first token is the encoder's latency, while time to each additional token is decoder's latency, where we assume a max decoded length specified below.
|
| 16 |
|
| 17 |
This is based on the implementation of Whisper-Small-Quantized found [here](https://github.com/huggingface/transformers/tree/v4.42.3/src/transformers/models/whisper).
|
| 18 |
-
This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/
|
| 19 |
|
| 20 |
Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) to run these models on a hosted Qualcomm® device.
|
| 21 |
|
|
@@ -28,40 +28,40 @@ Below are pre-exported model assets ready for deployment.
|
|
| 28 |
|
| 29 |
| Runtime | Precision | Chipset | SDK Versions | Download |
|
| 30 |
|---|---|---|---|---|
|
| 31 |
-
| PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.
|
| 32 |
-
| PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® X2 Elite | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.
|
| 33 |
-
| PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® X Elite | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.
|
| 34 |
-
| PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® 8 Gen 3 Mobile | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.
|
| 35 |
-
| PRECOMPILED_QNN_ONNX | w8a16 | Qualcomm® QCS8550 (Proxy) | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.
|
| 36 |
-
| PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.
|
| 37 |
-
| PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® 7 Gen 4 Mobile | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.
|
| 38 |
-
| PRECOMPILED_QNN_ONNX | w8a16 | Qualcomm® QCM6690 | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.
|
| 39 |
-
| PRECOMPILED_QNN_ONNX | w8a16 | Qualcomm® QCS9075 | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.
|
| 40 |
-
| QNN_CONTEXT_BINARY | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.
|
| 41 |
-
| QNN_CONTEXT_BINARY | w8a16 | Snapdragon® X2 Elite | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.
|
| 42 |
-
| QNN_CONTEXT_BINARY | w8a16 | Snapdragon® X Elite | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.
|
| 43 |
-
| QNN_CONTEXT_BINARY | w8a16 | Snapdragon® 8 Gen 3 Mobile | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.
|
| 44 |
-
| QNN_CONTEXT_BINARY | w8a16 | Qualcomm® QCS8550 (Proxy) | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.
|
| 45 |
-
| QNN_CONTEXT_BINARY | w8a16 | Qualcomm® SA8775P | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.
|
| 46 |
-
| QNN_CONTEXT_BINARY | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.
|
| 47 |
-
| QNN_CONTEXT_BINARY | w8a16 | Snapdragon® 7 Gen 4 Mobile | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.
|
| 48 |
-
| QNN_CONTEXT_BINARY | w8a16 | Qualcomm® SA7255P | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.
|
| 49 |
-
| QNN_CONTEXT_BINARY | w8a16 | Qualcomm® QCM6690 | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.
|
| 50 |
-
| QNN_CONTEXT_BINARY | w8a16 | Qualcomm® QCS9075 | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.
|
| 51 |
|
| 52 |
For more device-specific assets and performance metrics, visit **[Whisper-Small-Quantized on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/whisper_small_quantized)**.
|
| 53 |
|
| 54 |
|
| 55 |
### Option 2: Export with Custom Configurations
|
| 56 |
|
| 57 |
-
Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/
|
| 58 |
- Custom weights (e.g., fine-tuned checkpoints)
|
| 59 |
- Custom input shapes
|
| 60 |
- Target device and runtime configurations
|
| 61 |
|
| 62 |
This option is ideal if you need to customize the model beyond the default configuration provided here.
|
| 63 |
|
| 64 |
-
See our repository for [Whisper-Small-Quantized on GitHub](https://github.com/qualcomm/ai-hub-models/
|
| 65 |
|
| 66 |
## Model Details
|
| 67 |
|
|
@@ -75,48 +75,48 @@ See our repository for [Whisper-Small-Quantized on GitHub](https://github.com/qu
|
|
| 75 |
## Performance Summary
|
| 76 |
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|
| 77 |
|---|---|---|---|---|---|---
|
| 78 |
-
|
|
| 79 |
-
|
|
| 80 |
-
|
|
| 81 |
-
|
|
| 82 |
-
|
|
| 83 |
-
|
|
| 84 |
-
|
|
| 85 |
-
|
|
| 86 |
-
|
|
| 87 |
-
|
|
| 88 |
-
|
|
| 89 |
-
|
|
| 90 |
-
|
|
| 91 |
-
|
|
| 92 |
-
|
|
| 93 |
-
|
|
| 94 |
-
|
|
| 95 |
-
|
|
| 96 |
-
|
|
| 97 |
-
|
|
| 98 |
-
|
|
| 99 |
-
|
|
| 100 |
-
|
|
| 101 |
-
|
|
| 102 |
-
|
|
| 103 |
-
|
|
| 104 |
-
|
|
| 105 |
-
|
|
| 106 |
-
|
|
| 107 |
-
|
|
| 108 |
-
|
|
| 109 |
-
|
|
| 110 |
-
|
|
| 111 |
-
|
|
| 112 |
-
|
|
| 113 |
-
|
|
| 114 |
-
|
|
| 115 |
-
|
|
| 116 |
-
|
|
| 117 |
-
|
|
| 118 |
-
|
|
| 119 |
-
|
|
| 120 |
|
| 121 |
## License
|
| 122 |
* The license for the original implementation of Whisper-Small-Quantized can be found
|
|
|
|
| 15 |
We have applied w8a16 quantization to significantly enhance performance and efficiency. HuggingFace Whisper-Small ASR (Automatic Speech Recognition) model is a state-of-the-art system designed for transcribing spoken language into written text. This model is based on the transformer architecture and has been optimized for edge inference by replacing Multi-Head Attention (MHA) with Single-Head Attention (SHA) and linear layers with convolutional (conv) layers. It exhibits robust performance in realistic, noisy environments, making it highly reliable for real-world applications. Specifically, it excels in long-form transcription, capable of accurately transcribing audio clips up to 30 seconds long. Time to the first token is the encoder's latency, while time to each additional token is decoder's latency, where we assume a max decoded length specified below.
|
| 16 |
|
| 17 |
This is based on the implementation of Whisper-Small-Quantized found [here](https://github.com/huggingface/transformers/tree/v4.42.3/src/transformers/models/whisper).
|
| 18 |
+
This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/qai_hub_models/models/whisper_small_quantized) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).
|
| 19 |
|
| 20 |
Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) to run these models on a hosted Qualcomm® device.
|
| 21 |
|
|
|
|
| 28 |
|
| 29 |
| Runtime | Precision | Chipset | SDK Versions | Download |
|
| 30 |
|---|---|---|---|---|
|
| 31 |
+
| PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-precompiled_qnn_onnx-w8a16-qualcomm_snapdragon_8_elite_gen5.zip)
|
| 32 |
+
| PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® X2 Elite | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-precompiled_qnn_onnx-w8a16-qualcomm_snapdragon_x2_elite.zip)
|
| 33 |
+
| PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® X Elite | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-precompiled_qnn_onnx-w8a16-qualcomm_snapdragon_x_elite.zip)
|
| 34 |
+
| PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® 8 Gen 3 Mobile | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-precompiled_qnn_onnx-w8a16-qualcomm_snapdragon_8gen3.zip)
|
| 35 |
+
| PRECOMPILED_QNN_ONNX | w8a16 | Qualcomm® QCS8550 (Proxy) | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-precompiled_qnn_onnx-w8a16-qualcomm_qcs8550_proxy.zip)
|
| 36 |
+
| PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-precompiled_qnn_onnx-w8a16-qualcomm_snapdragon_8_elite_for_galaxy.zip)
|
| 37 |
+
| PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® 7 Gen 4 Mobile | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-precompiled_qnn_onnx-w8a16-qualcomm_snapdragon_7gen4.zip)
|
| 38 |
+
| PRECOMPILED_QNN_ONNX | w8a16 | Qualcomm® QCM6690 | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-precompiled_qnn_onnx-w8a16-qualcomm_qcm6690.zip)
|
| 39 |
+
| PRECOMPILED_QNN_ONNX | w8a16 | Qualcomm® QCS9075 | QAIRT 2.42, ONNX Runtime 1.24.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-precompiled_qnn_onnx-w8a16-qualcomm_qcs9075.zip)
|
| 40 |
+
| QNN_CONTEXT_BINARY | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-qnn_context_binary-w8a16-qualcomm_snapdragon_8_elite_gen5.zip)
|
| 41 |
+
| QNN_CONTEXT_BINARY | w8a16 | Snapdragon® X2 Elite | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-qnn_context_binary-w8a16-qualcomm_snapdragon_x2_elite.zip)
|
| 42 |
+
| QNN_CONTEXT_BINARY | w8a16 | Snapdragon® X Elite | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-qnn_context_binary-w8a16-qualcomm_snapdragon_x_elite.zip)
|
| 43 |
+
| QNN_CONTEXT_BINARY | w8a16 | Snapdragon® 8 Gen 3 Mobile | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-qnn_context_binary-w8a16-qualcomm_snapdragon_8gen3.zip)
|
| 44 |
+
| QNN_CONTEXT_BINARY | w8a16 | Qualcomm® QCS8550 (Proxy) | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-qnn_context_binary-w8a16-qualcomm_qcs8550_proxy.zip)
|
| 45 |
+
| QNN_CONTEXT_BINARY | w8a16 | Qualcomm® SA8775P | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-qnn_context_binary-w8a16-qualcomm_sa8775p.zip)
|
| 46 |
+
| QNN_CONTEXT_BINARY | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-qnn_context_binary-w8a16-qualcomm_snapdragon_8_elite_for_galaxy.zip)
|
| 47 |
+
| QNN_CONTEXT_BINARY | w8a16 | Snapdragon® 7 Gen 4 Mobile | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-qnn_context_binary-w8a16-qualcomm_snapdragon_7gen4.zip)
|
| 48 |
+
| QNN_CONTEXT_BINARY | w8a16 | Qualcomm® SA7255P | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-qnn_context_binary-w8a16-qualcomm_sa7255p.zip)
|
| 49 |
+
| QNN_CONTEXT_BINARY | w8a16 | Qualcomm® QCM6690 | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-qnn_context_binary-w8a16-qualcomm_qcm6690.zip)
|
| 50 |
+
| QNN_CONTEXT_BINARY | w8a16 | Qualcomm® QCS9075 | QAIRT 2.43 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-qnn_context_binary-w8a16-qualcomm_qcs9075.zip)
|
| 51 |
|
| 52 |
For more device-specific assets and performance metrics, visit **[Whisper-Small-Quantized on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/whisper_small_quantized)**.
|
| 53 |
|
| 54 |
|
| 55 |
### Option 2: Export with Custom Configurations
|
| 56 |
|
| 57 |
+
Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/qai_hub_models/models/whisper_small_quantized) Python library to compile and export the model with your own:
|
| 58 |
- Custom weights (e.g., fine-tuned checkpoints)
|
| 59 |
- Custom input shapes
|
| 60 |
- Target device and runtime configurations
|
| 61 |
|
| 62 |
This option is ideal if you need to customize the model beyond the default configuration provided here.
|
| 63 |
|
| 64 |
+
See our repository for [Whisper-Small-Quantized on GitHub](https://github.com/qualcomm/ai-hub-models/blob/main/qai_hub_models/models/whisper_small_quantized) for usage instructions.
|
| 65 |
|
| 66 |
## Model Details
|
| 67 |
|
|
|
|
| 75 |
## Performance Summary
|
| 76 |
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|
| 77 |
|---|---|---|---|---|---|---
|
| 78 |
+
| decoder | PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 4.017 ms | 36 - 46 MB | NPU
|
| 79 |
+
| decoder | PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® X2 Elite | 3.825 ms | 185 - 185 MB | NPU
|
| 80 |
+
| decoder | PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® X Elite | 7.953 ms | 185 - 185 MB | NPU
|
| 81 |
+
| decoder | PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® 8 Gen 3 Mobile | 6.47 ms | 39 - 48 MB | NPU
|
| 82 |
+
| decoder | PRECOMPILED_QNN_ONNX | w8a16 | Qualcomm® QCS8550 (Proxy) | 8.304 ms | 27 - 29 MB | NPU
|
| 83 |
+
| decoder | PRECOMPILED_QNN_ONNX | w8a16 | Qualcomm® QCS9075 | 9.145 ms | 24 - 57 MB | NPU
|
| 84 |
+
| decoder | PRECOMPILED_QNN_ONNX | w8a16 | Qualcomm® QCM6690 | 32.401 ms | 28 - 37 MB | NPU
|
| 85 |
+
| decoder | PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 4.776 ms | 25 - 37 MB | NPU
|
| 86 |
+
| decoder | PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® 7 Gen 4 Mobile | 10.88 ms | 30 - 36 MB | NPU
|
| 87 |
+
| decoder | QNN_CONTEXT_BINARY | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 3.967 ms | 30 - 41 MB | NPU
|
| 88 |
+
| decoder | QNN_CONTEXT_BINARY | w8a16 | Snapdragon® X2 Elite | 4.249 ms | 30 - 30 MB | NPU
|
| 89 |
+
| decoder | QNN_CONTEXT_BINARY | w8a16 | Snapdragon® X Elite | 7.557 ms | 30 - 30 MB | NPU
|
| 90 |
+
| decoder | QNN_CONTEXT_BINARY | w8a16 | Snapdragon® 8 Gen 3 Mobile | 6.203 ms | 19 - 27 MB | NPU
|
| 91 |
+
| decoder | QNN_CONTEXT_BINARY | w8a16 | Qualcomm® QCS8275 (Proxy) | 13.659 ms | 12 - 19 MB | NPU
|
| 92 |
+
| decoder | QNN_CONTEXT_BINARY | w8a16 | Qualcomm® QCS8550 (Proxy) | 8.187 ms | 30 - 32 MB | NPU
|
| 93 |
+
| decoder | QNN_CONTEXT_BINARY | w8a16 | Qualcomm® SA8775P | 27.56 ms | 19 - 26 MB | NPU
|
| 94 |
+
| decoder | QNN_CONTEXT_BINARY | w8a16 | Qualcomm® QCS9075 | 8.936 ms | 25 - 60 MB | NPU
|
| 95 |
+
| decoder | QNN_CONTEXT_BINARY | w8a16 | Qualcomm® QCM6690 | 30.781 ms | 30 - 37 MB | NPU
|
| 96 |
+
| decoder | QNN_CONTEXT_BINARY | w8a16 | Qualcomm® SA7255P | 13.659 ms | 12 - 19 MB | NPU
|
| 97 |
+
| decoder | QNN_CONTEXT_BINARY | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 4.793 ms | 28 - 41 MB | NPU
|
| 98 |
+
| decoder | QNN_CONTEXT_BINARY | w8a16 | Snapdragon® 7 Gen 4 Mobile | 10.907 ms | 29 - 36 MB | NPU
|
| 99 |
+
| encoder | PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 174.548 ms | 63 - 73 MB | NPU
|
| 100 |
+
| encoder | PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® X2 Elite | 155.383 ms | 127 - 127 MB | NPU
|
| 101 |
+
| encoder | PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® X Elite | 267.039 ms | 127 - 127 MB | NPU
|
| 102 |
+
| encoder | PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® 8 Gen 3 Mobile | 240.522 ms | 63 - 74 MB | NPU
|
| 103 |
+
| encoder | PRECOMPILED_QNN_ONNX | w8a16 | Qualcomm® QCS8550 (Proxy) | 386.76 ms | 55 - 58 MB | NPU
|
| 104 |
+
| encoder | PRECOMPILED_QNN_ONNX | w8a16 | Qualcomm® QCS9075 | 255.083 ms | 63 - 66 MB | NPU
|
| 105 |
+
| encoder | PRECOMPILED_QNN_ONNX | w8a16 | Qualcomm® QCM6690 | 4338.57 ms | 1 - 12 MB | NPU
|
| 106 |
+
| encoder | PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 200.696 ms | 64 - 75 MB | NPU
|
| 107 |
+
| encoder | PRECOMPILED_QNN_ONNX | w8a16 | Snapdragon® 7 Gen 4 Mobile | 457.646 ms | 56 - 62 MB | NPU
|
| 108 |
+
| encoder | QNN_CONTEXT_BINARY | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 173.204 ms | 1 - 11 MB | NPU
|
| 109 |
+
| encoder | QNN_CONTEXT_BINARY | w8a16 | Snapdragon® X2 Elite | 154.118 ms | 0 - 0 MB | NPU
|
| 110 |
+
| encoder | QNN_CONTEXT_BINARY | w8a16 | Snapdragon® X Elite | 295.657 ms | 0 - 0 MB | NPU
|
| 111 |
+
| encoder | QNN_CONTEXT_BINARY | w8a16 | Snapdragon® 8 Gen 3 Mobile | 267.017 ms | 3 - 10 MB | NPU
|
| 112 |
+
| encoder | QNN_CONTEXT_BINARY | w8a16 | Qualcomm® QCS8275 (Proxy) | 515.755 ms | 1 - 9 MB | NPU
|
| 113 |
+
| encoder | QNN_CONTEXT_BINARY | w8a16 | Qualcomm® QCS8550 (Proxy) | 366.478 ms | 1 - 2 MB | NPU
|
| 114 |
+
| encoder | QNN_CONTEXT_BINARY | w8a16 | Qualcomm® SA8775P | 310.162 ms | 0 - 9 MB | NPU
|
| 115 |
+
| encoder | QNN_CONTEXT_BINARY | w8a16 | Qualcomm® QCS9075 | 292.347 ms | 0 - 29 MB | NPU
|
| 116 |
+
| encoder | QNN_CONTEXT_BINARY | w8a16 | Qualcomm® QCM6690 | 4026.592 ms | 0 - 7 MB | NPU
|
| 117 |
+
| encoder | QNN_CONTEXT_BINARY | w8a16 | Qualcomm® SA7255P | 515.755 ms | 1 - 9 MB | NPU
|
| 118 |
+
| encoder | QNN_CONTEXT_BINARY | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 223.174 ms | 1 - 10 MB | NPU
|
| 119 |
+
| encoder | QNN_CONTEXT_BINARY | w8a16 | Snapdragon® 7 Gen 4 Mobile | 471.739 ms | 1 - 7 MB | NPU
|
| 120 |
|
| 121 |
## License
|
| 122 |
* The license for the original implementation of Whisper-Small-Quantized can be found
|
release_assets.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"version":"0.50.0","precisions":{"w8a16":{"chipset_assets":{"qualcomm-snapdragon-8gen3":{"qnn_context_binary":{"tool_versions":{"qairt":"2.43.0.260127150333_193827"},"download_url":"https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-qnn_context_binary-w8a16-qualcomm_snapdragon_8gen3.zip"},"precompiled_qnn_onnx":{"tool_versions":{"qairt":"2.42.0.251225135753_193295","onnx_runtime":"1.24.1"},"download_url":"https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-precompiled_qnn_onnx-w8a16-qualcomm_snapdragon_8gen3.zip"}},"qualcomm-snapdragon-8-elite-for-galaxy":{"qnn_context_binary":{"tool_versions":{"qairt":"2.43.0.260127150333_193827"},"download_url":"https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-qnn_context_binary-w8a16-qualcomm_snapdragon_8_elite_for_galaxy.zip"},"precompiled_qnn_onnx":{"tool_versions":{"qairt":"2.42.0.251225135753_193295","onnx_runtime":"1.24.1"},"download_url":"https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-precompiled_qnn_onnx-w8a16-qualcomm_snapdragon_8_elite_for_galaxy.zip"}},"qualcomm-snapdragon-7gen4":{"qnn_context_binary":{"tool_versions":{"qairt":"2.43.0.260127150333_193827"},"download_url":"https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-qnn_context_binary-w8a16-qualcomm_snapdragon_7gen4.zip"},"precompiled_qnn_onnx":{"tool_versions":{"qairt":"2.42.0.251225135753_193295","onnx_runtime":"1.24.1"},"download_url":"https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-precompiled_qnn_onnx-w8a16-qualcomm_snapdragon_7gen4.zip"}},"qualcomm-snapdragon-8-elite-gen5":{"qnn_context_binary":{"tool_versions":{"qairt":"2.43.0.260127150333_193827"},"download_url":"https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-qnn_context_binary-w8a16-qualcomm_snapdragon_8_elite_gen5.zip"},"precompiled_qnn_onnx":{"tool_versions":{"qairt":"2.42.0.251225135753_193295","onnx_runtime":"1.24.1"},"download_url":"https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-precompiled_qnn_onnx-w8a16-qualcomm_snapdragon_8_elite_gen5.zip"}},"qualcomm-snapdragon-x-elite":{"qnn_context_binary":{"tool_versions":{"qairt":"2.43.0.260127150333_193827"},"download_url":"https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-qnn_context_binary-w8a16-qualcomm_snapdragon_x_elite.zip"},"precompiled_qnn_onnx":{"tool_versions":{"qairt":"2.42.0.251225135753_193295","onnx_runtime":"1.24.1"},"download_url":"https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-precompiled_qnn_onnx-w8a16-qualcomm_snapdragon_x_elite.zip"}},"qualcomm-snapdragon-x2-elite":{"qnn_context_binary":{"tool_versions":{"qairt":"2.43.0.260127150333_193827"},"download_url":"https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-qnn_context_binary-w8a16-qualcomm_snapdragon_x2_elite.zip"},"precompiled_qnn_onnx":{"tool_versions":{"qairt":"2.42.0.251225135753_193295","onnx_runtime":"1.24.1"},"download_url":"https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-precompiled_qnn_onnx-w8a16-qualcomm_snapdragon_x2_elite.zip"}},"qualcomm-sa7255p":{"qnn_context_binary":{"tool_versions":{"qairt":"2.43.0.260127150333_193827"},"download_url":"https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-qnn_context_binary-w8a16-qualcomm_sa7255p.zip"}},"qualcomm-sa8775p":{"qnn_context_binary":{"tool_versions":{"qairt":"2.43.0.260127150333_193827"},"download_url":"https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-qnn_context_binary-w8a16-qualcomm_sa8775p.zip"}},"qualcomm-qcm6690":{"qnn_context_binary":{"tool_versions":{"qairt":"2.43.0.260127150333_193827"},"download_url":"https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-qnn_context_binary-w8a16-qualcomm_qcm6690.zip"},"precompiled_qnn_onnx":{"tool_versions":{"qairt":"2.42.0.251225135753_193295","onnx_runtime":"1.24.1"},"download_url":"https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-precompiled_qnn_onnx-w8a16-qualcomm_qcm6690.zip"}},"qualcomm-qcs8550-proxy":{"qnn_context_binary":{"tool_versions":{"qairt":"2.43.0.260127150333_193827"},"download_url":"https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-qnn_context_binary-w8a16-qualcomm_qcs8550_proxy.zip"},"precompiled_qnn_onnx":{"tool_versions":{"qairt":"2.42.0.251225135753_193295","onnx_runtime":"1.24.1"},"download_url":"https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-precompiled_qnn_onnx-w8a16-qualcomm_qcs8550_proxy.zip"}},"qualcomm-qcs9075":{"qnn_context_binary":{"tool_versions":{"qairt":"2.43.0.260127150333_193827"},"download_url":"https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-qnn_context_binary-w8a16-qualcomm_qcs9075.zip"},"precompiled_qnn_onnx":{"tool_versions":{"qairt":"2.42.0.251225135753_193295","onnx_runtime":"1.24.1"},"download_url":"https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/whisper_small_quantized/releases/v0.50.0/whisper_small_quantized-precompiled_qnn_onnx-w8a16-qualcomm_qcs9075.zip"}}}}}}
|